Page 2 of 2 FirstFirst 12
Results 11 to 17 of 17

Thread: 25ms vs 50ms with vixen to FPP

  1. #11
    Join Date
    Nov 2016
    Location
    North Vancouver, BC, Canada
    Posts
    1,258
    Post Thanks / Like

    Default Re: 25ms vs 50ms with vixen to FPP

    Quote Originally Posted by MartinMueller2003 View Post
    It is a perception thing. Many people run their show at 50ms refresh rate which is 1360 WS281x pixels. If you feed the ESP with 25ms data, your 800 pixels will play at ~35ms. Would anyone notice?
    Bingo! Exactly it! I really don't care if its a little lower than that 40fps target on playback because it's likely to be imperceptible between 35 and 40fps. Even if it got Down to 30fps, I'd still take that performance increase over 20fps. As long as it still functions as intended and stays in sync, it's a win in my books.

    And if anyone is wondering why I'd be so anal about trying to get the faster framerate other than it looking smoother, prime example: The nutrocker! Lol

  2. #12
    Join Date
    Dec 2015
    Location
    Farnborough, UK
    Posts
    39
    Post Thanks / Like

    Default Re: 25ms vs 50ms with vixen to FPP

    Funny I was thinking and questioned the same thing in another thread - would love to know if there is any visible difference between 25/50ms - I've always run 50ms as well

  3. #13
    Join Date
    Dec 2012
    Location
    Newtown CT
    Posts
    5,650
    Post Thanks / Like

    Default Re: 25ms vs 50ms with vixen to FPP

    Quote Originally Posted by Mat-Moo View Post
    Funny I was thinking and questioned the same thing in another thread - would love to know if there is any visible difference between 25/50ms - I've always run 50ms as well
    I have rendered my show in both 50ms and 25ms modes. To me there is a significant difference in the smoothness of many effects. Now I am sure there is a bit of the observer effect here (I am the one who set the speed) so a better test would have been to have someone else run the different files and then see if I can tell the difference.


    2021 New Tune-to sign. New 40x27 matrix at the house. Retiring the Strip based matrix. Updating some of the old window frames to the new house. Adding two new songs.
    2020 Full sized show reworked for the new location. Only adding (famous last words) 13 RBLs that I finally got converted to using pixels
    2019 - Just moved into a new home (yet another change of plans). Will be dim but not dark. Too much to do at the new place to leave time for a show. Dim show (3000 pixels) had regular visits most nights.
    https://www.youtube.com/user/MartinMueller2003

  4. #14
    Join Date
    Nov 2016
    Location
    North Vancouver, BC, Canada
    Posts
    1,258
    Post Thanks / Like

    Default Re: 25ms vs 50ms with vixen to FPP

    Quote Originally Posted by Mat-Moo View Post
    Funny I was thinking and questioned the same thing in another thread - would love to know if there is any visible difference between 25/50ms - I've always run 50ms as well

    Oh there absolutely is! Without a doubt. If someone can't notice a difference I'd question their perception lol. But it's degrees of difference based very much on the type of effect and movement is when it's most noticeable. So certain effects you may not be quite sure, but many, especially with fast motion, location based etc it's noticeably smoother. To be clear, I've generally been fine with 50ms, it's not "bad" at all. Although the way I sequence it felt like it wasn't quite as responsive as intended. So if this works, I'll be pretty happy. Once I have it up and running and tested, I'll record some clips with both versions to compare

  5. #15
    Join Date
    Oct 2014
    Location
    Sauk City, WI USA
    Posts
    2,231
    Post Thanks / Like

    Default Re: 25ms vs 50ms with vixen to FPP

    So here is the problem with 800 pixels at 25ms. According to Martin, the frame will be 35ms long.

    First packet. no problem. data gets passed down the string, the timeout comes along and the lights come on. The last ones, pretty close to 25ms but not quite.
    Second packet. data gets passed down the string, the timeout comes along and the lights come on. We are now at 70ms in real time but it should have been done in 50ms.
    Third packet. ... we are now at 105ms and should be at 75.
    Fourth packet. ... we are now at 140ms real time and should be at 100.
    Sync packet!
    Oops, we are off by 40ms
    skip next (two?) packet and go on to next. Did you notice the missing packet?
    Once you get to the 10 frame sync, you will be off 100ms or 4 data packets. Will you notice that "glitch" every 250ms? (that is 4 dropped frames - 4 times a second).

  6. #16
    Join Date
    Nov 2011
    Location
    Chicago - Southwest Suburbs
    Posts
    8,530
    Post Thanks / Like

    Default Re: 25ms vs 50ms with vixen to FPP

    Mike's example is one potential behavior. You really don't know how the controllers handle their buffers under the hood.
    I've been in the assembly code of the sandevices controllers and know that it uses the same buffer for input and output. You'd probably crash the 68x controllers by feeding it data faster than it could empty the buffer. After it receives a universe of data, it then starts writing it out to the pixels, on a per string basis. (actually by bank on the older firmware) I'm not sure off hand if it would reenter the sending routine if it receives another frame before it's finished pumping out the first one. I think it probably wouldn't, but rather would just miss that call, and you'd actually drop every other update of the pixel string, giving you half the fps than you were shooting for anyhow. But since the next frame arrived in the buffer while it's streaming it out, the first part of the sting would use the data from the first frame, and the later part of the string would use the data that came in in the meantime.
    Different controllers will handle that differently of course, but my point is that you don't really know how it's going to behave.

  7. #17
    Join Date
    Dec 2012
    Location
    Newtown CT
    Posts
    5,650
    Post Thanks / Like

    Default Re: 25ms vs 50ms with vixen to FPP

    In the ESP I used a single buffer between input and output but I made the processing asynchronous. The output processing has no knowledge about the input processing. It just sends the data at the fastest frame rate possible always starting from the first byte in the buffer. The input processing moves data into the buffer in a way that the locations in the buffer align with a specific channel. The data can arrive at any rate it wants and it will not impact output processing.


    2021 New Tune-to sign. New 40x27 matrix at the house. Retiring the Strip based matrix. Updating some of the old window frames to the new house. Adding two new songs.
    2020 Full sized show reworked for the new location. Only adding (famous last words) 13 RBLs that I finally got converted to using pixels
    2019 - Just moved into a new home (yet another change of plans). Will be dim but not dark. Too much to do at the new place to leave time for a show. Dim show (3000 pixels) had regular visits most nights.
    https://www.youtube.com/user/MartinMueller2003

Page 2 of 2 FirstFirst 12

Bookmarks

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •